Recent investigations of the history of Chinese have given new support to the view that sound change diffuses gradually across the lexicon. Yet instrumental studies of sound change in progress support the Neogrammarian position that change affects all words that include the sound according to their phonetic environment. The paradox can be resolved by distinguishing abstract phonological change from change in low-level output rules. Both types of rules can be observed in recent studies of sound change in progress in Philadelphia: the lexical split of short a shows lexical diffusion in progress, while raising, lowering, fronting, and backing rules show Neogrammarian regularity. A review of the literature on completed changes and other changes in progress tends to support the relevance of a hierarchy of abstractness in determining the nature of the transition from one stage to the other.

A number of Australian Aboriginal languages have syntactic mechanisms which can be analysed as instances of ‘switch-reference’. These languages have verb suffixes indicating whether or not the subjects of syntactically-related main and subordinate clauses are referentially the same or different. The distribution of switch-reference in various subordinate clause types is examined here, and its formal and functional characteristics are described in detail for a number of language groups. The languages which have switch-reference are all geographically adjacent, but are not obviously related genetically. When the geographical spread of certain formal features such as the association between locative case and some ‘relative clause’ types is discussed, a number of general conclusions can be drawn, among them the suggestion that the syntactic mechanism of switch-reference has been subject to indirect functional diffusion between language groups. It is further suggested that careful attention to the precise nature of syntactic similarities is necessary if any attempts to reconstruct the syntax of proto-languages in Australia are to be undertaken. This paper provides further evidence of the importance of linguistic diffusion in Australia, and raises a number of issues relevant to studies of areal features elsewhere in the world.

The languages of Arnhem Land, Australia, form a contiguous block and Sprachbund, despite a sharp genetic boundary. Thus in Ngandi and Ritharngu, which belong to different major genetic divisions, about half the total vocabulary is shared, virtually all because of recent diffusion. ‘Core’ vocabulary and highly inflected stems (verbs) are among the borrowed items, but linguistic structural factors affect the extent of diffusion. Extensive over-all borrowing is attributed to local demographic and cultural features: small size of ethnolinguistic group, absence of strong tribal organization, and extensive cross-linguistic marriage.

The mental basis of linguistic intuitions is obscure, as regards their relationship both to other aspects of language behavior, such as speaking and listening, and to an hypothesized epistemological structure, such as a ‘grammar’. In the present study, we show that experimentally manipulated differences in mental state can systematically alter the linguistic intuitions which speakers render about sentences. These results indicate that the processes underlying intuitions cannot be ignored when they are used as empirical data to test grammatical theories.

‘Inner perception ... constitutes the ultimate and indispensable precondition for the other ... sources of knowledge. Consequently ... inner perception constitutes the very foundation upon which the science of psychology is erected’ (Brentano [1874] 1973:43) ... ‘If anyone were to mount a skeptical attack against this ultimate foundation of cognition, he would find no other foundation upon which to erect an edifice of knowledge. Thus, there is no need to justify our confidence in inner perception. What is clearly needed instead is a theory about the relation between such perception and its object ... such a theory is no longer possible if perception and object are separated into two distinct mental acts, of which the one would only be an effect of the other’ (ibid., 140).

Kay & McDaniel 1978 have recently proposed a new method of distinguishing between basic and non-basic color terms, based on fuzzy set theory (Zadeh 1965). In the present paper, we report two empirical studies that test the effectiveness of the fuzzy set method for identifying basic color categories. It is found that the method fails to discriminate between basic and non-basic categories. It is argued on theoretical grounds that the method proposed by Kay & McDaniel cannot distinguish between basic and non-basic colors because its primary premise, that the two types of categories display different characteristic membership functions, is false. Some general conclusions about the structure of categories are offered.

Play languages involving substitution, repetition, or re-arrangement of the sound units or words of some primary language are common, especially among children. In a great majority of these systems, the rules are few in number (not more than four or five) and simple in operation. However, this paper describes a highly complex secret language called Spaka that was developed more than twenty years ago by one of the authors, and is still used as the principal means of oral communication between her and her sister. The rules of Spaka alter certain syntactic and morphological aspects of English surface structure; in addition, they produce elaborate changes in English phonetic representations.

This paper critically examines Jerrold Katz's position on meaning, in particular his view that a semantic theory can be constructed which is free of information derived from pragmatics and speakers' belief systems. It is shown that, if such a semantic theory exists, it either excludes many fundamental phenomena which are normally thought of as semantic, or else misses linguistically significant generalizations. It is then argued that, even if Katz's particular version of semantics is rejected, this does not render an intentionalist semantic theory impossible.